Citation Recommendation Based On Community Merging And Time Effect

In the information age,cloud computing,5G communication technology and other application technologies are shining,and users are enjoying the convenience brought by information technology.The means for users to obtain information has changed from paper newspapers to mobile phones and tablets,and has shrunk from square meters to square centimeters.Users no longer worry about the way to obtain information,but need to choose the information they are interested in reading.People have found themselves lost in the whirlpool of the information abyss while enjoying an excess of information needs.The birth of the recommendation system marks the opening of a new chapter of personalized information service for user services.It can recommend users' potential needs of interest according to users' behavior and interest,and has been actively developed and put into practical application in various fields.Different organizations have been increasing the construction of electronic document resources,people are very convenient to use the document resources provided by the organization on the network,but inevitably encounter some troubles.For example,it takes time and effort for users to find an article that fits the research direction,and the articles in the recommended list are unsatisfactory.In citation recommendation research,accurately obtaining a target article has become a key research topic,with the purpose of saving the time of preparatory work and effectively improving the quality and efficiency of researchers' writing.The citation recommendation system is different from the common shopping recommendation system.Research workers are the main service objects of citation recommendation,and there are many influencing factors for recommending academic articles,such as citation category,research field,author popularity,etc.In terms of citation recommendation research,the research time in foreign countries is much longer than that in China,but the optimization of recommendation algorithm is basically based on algorithm improvement.Most recommendation algorithms do not fully consider the similarity of internal semantic information and the time of publication of papers.Based on this kind of problem,referencing to the classical Path Sim algorithm(Top-K similarity search algorithm based on meta-path in heterogeneous information graph),a new similarity measurement algorithm CMTE-Pathsim(Pathsim Based on Community Merging and Time Effect)is proposed,and a new dividing mode of heterogeneous information network is designed.The unrelated nodes in the heterogeneous information network can be excluded to make the similar nodes form smaller heterogeneous information network.Time factor is introduced into the similarity calculation to reorder the recommendation list.After comparing and analyzing with P-Page Rank,Path Sim and Simrank on the evaluation standards of Recall@k,Precision@k and F1.The proposed algorithm produces a more reasonable and excellent recommendation list effect.